Big Blue Goes for the Big Win

IBM's director of R&D is shifting the tech giant's focusand making a few enormous bets

It's a snowy February day at IBM's office in Yorktown Heights, N.Y., and John E. Kelly III drops by a cramped conference room to talk about his plans for IBM Research. The organization is already considered one of the world's best corporate research labs. Yet Kelly, a 27-year IBM veteran who took over as research director in July, is planning surprisingly dramatic changes. "We have to do bolder things, bigger things," he says, speaking about his plans publicly for the first time. "If we don't fail a third of the time, we're not stretching enough. On the other hand, when we win, we need to win big."

What does Kelly have in mind? For starters, he's focusing on four top research priorities, rather than spreading investments too thin. The four bets are enormous, though. Each of the projects will get $100 million over the next two to three years, in hopes of generating at least $1 billion, each, in new revenue. The projects: inventing a successor to today's semiconductor, designing computers that process data much more efficiently, using math to solve complex business problems, and building massive clusters of computers that operate like a single machine—an approach called "cloud" computing. Central to the effort will be even more emphasis on basic scientific research, such as physics, chemistry, and math.

The other major change Kelly has in the works is overhauling the way IBM does research. Today the tech giant has eight research facilities with 3,200 scientists, and it hasn't opened a new one in a decade. Kelly foresees creating dozens of new joint ventures for research, which he calls "collaboratories," with countries, companies, and independent research outfits. One venture with Saudi Arabia, focusing on nanotechnology, was unveiled on Feb. 26. Kelly believes IBM needs to leverage its resources and learn from others. "The nature of research itself is changing," he says. "Great ideas are springing up everywhere, and we need to shift from focusing on large brick-and-mortar operations to having a much more collaborative outreach program."

He'll try to do all this without a sharp increase in spending. IBM doesn't break out its research budget, but Kelly says it will rise at the rate of IBM's revenues, in the single digits. The company spends more than $6 billion a year overall on research and development of new products.

Kelly's moves will be closely watched. IBM has been one of the few old-line companies to continue investing in basic research, as others have cut back. The once-vaunted Bell Labs, now owned by Alcatel-Lucent (ALU), has become a shadow of its former self. "IBM has been the only one with the staying power," says Tim Studt, editor-in-chief of R&D Magazine, which last year ranked IBM as the country's top R&D company. IBM has been the leading recipient of U.S. patents for 14 years running.

But Kelly knows IBM can't be complacent. Both Microsoft (MSFT) and Google (GOOG) are investing heavily in research. For the first time, Microsoft broke the top 10 for U.S. patent awards last year, coming in at No.6. "It's a race," says Kelly. "While others are trying to emulate us, we're going to skate ahead."

Kelly, the seventh director since IBM established its labs in 1945, has research in his blood. His father worked as a technician at General Electric (GE)'s lab in Niskayuna, N.Y., where Kelly would visit regularly as a boy and watch him work with vacuum tubes. Kelly got a PhD in materials engineering at Rensselaer Polytechnic Institute, just outside of Albany. At IBM, he ran one of the major research departments, in between stints in the chip division.

His collaborative research strategy emerged out of a dire situation he faced in 2003 as head of IBM's chip business. The company was suffering huge financial losses as the costs of keeping up with the latest technology were soaring.

Kelly took a gamble and set up research alliances with a handful of partners, including Sony Electronics (SNE) and Advanced Micro Devices (AMD), to share expenses and brainpower. The approach eventually paid off, as IBM's chip business returned to profitability and remained on the cutting edge of technology. "This is a smart strategy," says Studt. "You can't be the leader all by yourself anymore. The technology is just too complicated and expensive."

The just-unveiled deal with Saudi Arabia is an example of where Kelly is headed. The two sides plan to develop technologies for solar energy and water desalinization. The Saudis chose IBM because of its expertise. IBM scientists won a Nobel Prize in 1986 for nanotech breakthroughs, and they're leaders in developing new nanotech materials. "We want to work with the best in the field," says Turki Saud Al-Saud, vice-president for Research Institutes in King Abdulaziz City for Science & Technology.

Kelly expects the basic research that IBM does with partners to feed back into its own four high-priority projects. Nanotechnology, for instance, will be critical to inventing the successor to today's chips. Performance improvements have been slowing down in the current chip technology, and scientists expect chips of the future to be made with tiny switches built with individual atoms. As part of its effort, IBM announced on Feb. 22 that it had calculated how much force it takes to move a single atom. "If they can come up with a true, game-changing technology, they'll have a clear first-mover advantage and a huge business benefit," says Charles M. Vest, president of the National Academy of Engineering.

Another major research initiative with a big potential payoff is IBM's new computer reengineering project. IBM was at the forefront of redesigning microprocessors, the brains of computers, to include multiple parts, or cores, on a single piece of silicon. That makes it possible for the processors to run at lower power, with less heat. Now the company is designing chips with specialized cores that perform certain kinds of calculations more efficiently, plus software to take advantage of the new processing power. If the project works, it will help IBM strengthen its lead in high-margin server computers.

These are high-risk, high-reward ventures. "Big bets don't scare me at all," Kelly says. "I love this environment. I thrive on it."

    Before it's here, it's on the Bloomberg Terminal.